ML Projects
End-to-end data science, modeling, deployment, and evaluation workflows.
Physics PhD → Data Science, MLOps & Agentic AI Systems
Kian Maleki
I build reproducible machine learning systems, RAG/document-intelligence workflows, and quantitative models that turn complex data into reliable decisions.
My background combines physics-grade mathematical modeling, high-performance simulation, and practical ML engineering with Python, FastAPI, Docker, DVC, MLflow, Evidently, PyTorch, and Transformers.
Quality Gates
CI/CD workflows with pytest and model-performance thresholds before deployment.
Runtime Reduction
Optimized numerical simulations through algorithmic and parallel-computing improvements.
Targeted technical profile
Data Science, AI Systems & MLOps
Focused on roles in data science, machine learning engineering, AI agent systems, RAG, and document intelligence.
Machine Learning & Data Science
Generative AI & Agentic Systems
MLOps & Deployment
Quantitative & Scientific Computing
Selected work
Projects Matched to Data Science & AI Roles
Use the filters to focus on RAG, MLOps, APIs, document intelligence, or quantitative modeling.
RAG-Powered Knowledge Assistant
Built a local retrieval-augmented document question-answering system using Transformers and PyTorch, with sliding-window chunking to preserve semantic continuity across long documents.
- Designed context-aware chunking with configurable overlap.
- Used logit-based confidence scoring for answer extraction.
- Containerized the workflow with Docker for reproducible local execution.
Heart Disease Prediction: End-to-End MLOps Pipeline
Built a reproducible ML pipeline for heart disease prediction with data versioning, experiment tracking, model registry, automated tests, CI/CD, and drift monitoring.
- Tracked data and model artifacts with DVC and MLflow.
- Enforced CI checks with pytest and an accuracy threshold above 0.80.
- Added Evidently-based monitoring for data and concept drift.
SDF Document Information Extraction
Built a document intelligence pipeline to extract regulatory information from semi-structured SDF documents and convert unstructured text into structured datasets.
- Used Python, regex, NLP, and Pandas to identify and normalize key fields.
- Reduced manual review by converting document content into analysis-ready tables.
- Connected the work to document QA and RAG-style healthcare supply-chain automation.
House Price Prediction API Service
Built a REST API for real-time house price prediction using FastAPI, Pydantic validation, Scikit-Learn, and interactive API documentation.
- Designed schema-validated prediction endpoints.
- Exposed Swagger and ReDoc documentation for testing and adoption.
- Structured the project for reproducible deployment and extension.
Quantum Mechanical Modeling of Rare-Earth Materials
Developed theoretical and computational frameworks to study crystal fields, exchange and dipolar interactions, and noncollinear magnon dispersion in erbium oxide.
- Applied high-dimensional modeling to complex quantum materials.
- Published results in Physical Review B.
- Built publication-quality visualizations for theoretical results.
Noisy Tunneling Systems & Stochastic Modeling
Developed Monte Carlo simulations and signal-extraction workflows for stochastic experimental data in solid-state quantum device research.
- Modeled noisy tunneling behavior and extracted signal from stochastic data.
- Contributed to a Nano Letters publication and related patented device work.
- Reduced simulation runtime from days to under one hour through optimization.
Experience
Technical Background
Assistant Professor of Physics
Aug 2025 – PresentCreighton University
- Teach undergraduate physics courses and mentor teaching assistants.
- Develop interdisciplinary lab curriculum integrating ethical AI tools and chatbot-supported scientific workflows.
AI-Powered Document Intelligence Extern
2026Extern · Pfizer project context
- Built a functional document intelligence prototype for healthcare document processing.
- Developed retrieval-augmented workflows to improve document-based question answering.
Graduate Research Assistant
Aug 2018 – Aug 2025University of Iowa
- Developed high-dimensional modeling frameworks for quantum and spin systems.
- Optimized large-scale numerical simulations and extracted signal from noisy experimental data.
Graduate Research Assistant
Aug 2015 – Aug 2018Creighton University
- Built Python and Fortran simulations for physical systems, including N-body and molecular dynamics models.
- Parallelized scientific computing workloads for HPC resources including the Open Science Grid.
Open to targeted roles
Data Scientist · ML Engineer · AI / Agentic Systems
I am especially interested in roles involving RAG, document intelligence, model evaluation, MLOps, applied prediction systems, and physics-informed quantitative modeling.